Editorial introduction.

This special issue of the journal "Statistica Neerlandica" deals with various types of statistical and methodological problems encountered when applying multilevel and other types of random-coefficients models in social and behavioral science research. Six out of the seven papers were presented at the 2002 Autumn Meeting of the Social Science Section of the Netherlands Society for Statistics and Operations Research. In social science research, data are often hierarchically structured. Multilevel modeling techniques account for dependencies between observations within groups, as well as make it possible to investigate the importance of individual and group characteristics in the explanation of individual-level outcome variables.